From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1MOtzr-0001uf-Ah for qemu-devel@nongnu.org; Thu, 09 Jul 2009 09:49:11 -0400 Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43) id 1MOtzm-0001m6-MP for qemu-devel@nongnu.org; Thu, 09 Jul 2009 09:49:10 -0400 Received: from [199.232.76.173] (port=36507 helo=monty-python.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1MOtzk-0001ll-Bw for qemu-devel@nongnu.org; Thu, 09 Jul 2009 09:49:04 -0400 Received: from mx20.gnu.org ([199.232.41.8]:18381) by monty-python.gnu.org with esmtps (TLS-1.0:RSA_AES_256_CBC_SHA1:32) (Exim 4.60) (envelope-from ) id 1MOtzk-0001Hu-30 for qemu-devel@nongnu.org; Thu, 09 Jul 2009 09:49:04 -0400 Received: from mail.codesourcery.com ([65.74.133.4]) by mx20.gnu.org with esmtp (Exim 4.60) (envelope-from ) id 1MOtzi-0004rO-Vm for qemu-devel@nongnu.org; Thu, 09 Jul 2009 09:49:03 -0400 From: Paul Brook Subject: Re: [Qemu-devel] [PATCH 0/5] qdev: add driver class support. Date: Thu, 9 Jul 2009 14:48:59 +0100 References: <1247144544-8885-1-git-send-email-kraxel@redhat.com> <200907091419.11248.paul@codesourcery.com> <4A55F30C.9000502@redhat.com> In-Reply-To: <4A55F30C.9000502@redhat.com> MIME-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-15" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200907091448.59926.paul@codesourcery.com> List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Gerd Hoffmann Cc: qemu-devel@nongnu.org On Thursday 09 July 2009, Gerd Hoffmann wrote: > On 07/09/09 15:19, Paul Brook wrote: > > On Thursday 09 July 2009, Gerd Hoffmann wrote: > >> Hi, > >> > >> This patch series brings driver classes to qdev. > > > > I'm not sure this is a good idea. You should be able to figure it out > > from the device properties. > > I don't think this will work out in the general case. Some devices have > properties which can be used for that, i.e. network cards have a mac > address (no attribute yet but I expect it will come some day). Not > every device class has specific properties you can use to identify them > though. How would you identify a sound card for example? Ok, put it annother way: Why do you need to identify them? Why would libvirt care whether a device is (say) a sound card or a VGA adapter? In principle there's no reason why you shouldn't have many or none of both. IMHO the only reason we have the current grouping is because it's forced on us by the various commandline options which mix host and machine configuration. Paul